Linking education and hospital data in England: new publication demonstrates ECHILD linkage process and quality

Last week saw a new publication from the ECHILD team, with  ‘Linking education and hospital data in England: linkage process and quality’ published in the International Journal of Population Data Science (IJPDS).

The education and health records of nearly all (>95%) school children in England have, for the first time, been successfully linked to create an anonymised, securely held database for research. The ECHILD (Education and Child Health Insights from Linked Data) database will enable a significant improvement in the scale and depth of research into the relationships between health, education and social care across the lives of millions of children. However, researchers will need to be aware that linkage rates were not equal for groups of children from different backgrounds.

Records for all children born in four specific years and enrolled at state schools in England were studied and showed improvements in linkages rates over time – from 92% for those born in 1990/01 to 99% for children born in 2004/05. Advances in data quality through the introduction of the national Personal Demographics Service in 2004; the online system to register births from 2009; and better-quality data returned by hospitals, all contributed to these improvements.

However, researchers found lower linkage rates for children from ethnic minorities and those living in more deprived areas. This might reflect poor recording of non-traditional names, or more mobility in disadvantaged groups. First names, surnames, date of birth and postcodes were the identifiers used to link the children’s education records with their hospital records.

These differences in linkage accuracy are relatively small, but they highlight the need for health, welfare and educational services to continually improve data quality. And they show that poor recording of key data could have consequences for policy. For example, the data could underestimate health care needs in black or mixed-race groups because hospital records in these groups are less likely to link to education records than for white pupils.

Researchers addressing questions relating to ethnic minority or deprived groups need to consider, therefore, whether to adjust for missing data among these groups due to missed links. Statistical techniques can be used to account for missed links in analyses.

Overall, however, the high rate of successful linkages means the ECHILD database offers a vital resource for research that will help policymakers and services understand how health and educational needs and outcomes vary for children from different backgrounds and in different areas. By measuring differences between such groups of children, ECHILD can generate evidence for policymakers about inequalities in health and education, as services recover from the impact of Covid-19.

The database cannot be used by anyone to identify individuals. All personal information such as names and postcodes have been removed, and it can only be accessed by approved researchers in a secure environment at the Office for National Statistics.

Access the full paper on the IJDPS website.

Find out more about ECHILD on the ECHILD project website.